PCIe is experiencing its biggest change in 20 years, and may introduce optical transmission

The 1980s began, when computers integrated dozens of chips on motherboards and a large number of special expansion slots for adding additional cards. For the latter, one type mainly dominated the local scene: IBM's ISA bus (Industry Standard Architecture). Although improvements to the technology were not as successful in comparison, the system as a whole became ubiquitous within the industry.

By the time the next decade rolled around, faster processors helped drive the need for better-performing expansion buses, which ultimately resulted in two new formats—Intel's PCI bus (Peripheral Component Interconnect) and VLB (VESA Local Bus) from the Video Electronics Standards Association.

Both emerged at the same time in 1992, although PCI initially appeared to be the slower of the two, as it was designed to run at a fixed 33 MHz (later specification revisions did allow for 66 MHz, but consumer PCs never really supported this). In contrast, VLB runs at the same clock as the CPU's front-side bus (FSB), allowing VLB to reach 40 or 50 MHz, depending on the CPU.

However, it is not always stable at that rate, and the latency is worse than PCI. Typical VLB expansion slots are also much larger than PCI expansion slots. Despite these advantages, it took some time for PCI to gain traction in the motherboard industry, especially in the workstation and server markets.

At the time, home PC users generally didn't have many expansion cards, nor any that placed heavy demands on the bus. However, as the 3D graphics card industry took off, that changed, and the best graphics cards came with PCI connectors. As a result, motherboards began to favor the new bus over the old one. As these graphics accelerators grew in power and games took advantage of this, the limitations of the PCI bus became apparent.

Like ISA and VLB before it, PCI is a parallel data bus - meaning that all cards in a PCI expansion slot use the same bus and must take turns transmitting and receiving data. This can be problematic for graphics cards, as they can easily take over the bus. Intel solved this problem in 1997 by developing the Accelerated Graphics Port (AGP), which provided a dedicated PCI bus for graphics cards.

As the old millennium transitioned to the new, the need for a faster bus grew. Soon after Intel introduced the PCI bus, it formed a special interest group (PCI-SIG) to support motherboard and expansion card vendors in ensuring their hardware complied with the specification. By the early 2000s, this group consisted of hundreds of members, five of which (Compaq, Dell, HP, IBM, and Microsoft) worked with Intel to replace PCI.

The PCI-SIG, codenamed 3GIO (3rd Generation I/O), announced the fruits of its labor in April 2002, introducing a new technology called PCI Express. PCIe was born from this, and its full name is Peripheral Component Interconnect Express. Since then, PCIe has been continuously improved over time to adapt to the latest bandwidth requirements of modern computers.

Currently, there are more than 800 member companies that have joined PCI-SIG. Each company follows the specifications set by PCI-SIG and develops products with different functions but interoperable and communicative. The specifications that PCI-SIG has set include PCI, PCI-X and PCI Express.

The current PCI-SIG board of directors has also been expanded to include representatives from Intel, Dell, AMD, Qualcomm, HP, Synopsys, Agilent Technologies and Nvidia.